includes/parser/Parser_LinkHooks.php

   1 <?php
   2 /**
   3  * Modified version of the PHP parser with hooks for wiki links; experimental
   4  *
   5  * @file
   6  */
   7
   8 /**
   9  * Parser with LinkHooks experiment
  10  * @ingroup Parser
  11  */
  12 class Parser_LinkHooks extends Parser
  13 {
  14         /**
  15          * Update this version number when the ParserOutput format
  16          * changes in an incompatible way, so the parser cache
  17          * can automatically discard old data.
  18          */
  19         const VERSION = '1.6.4';
  20
  21         # Flags for Parser::setLinkHook
  22         # Also available as global constants from Defines.php
  23         const SLH_PATTERN = 1;
  24
  25         # Constants needed for external link processing
  26         # Everything except bracket, space, or control characters
  27         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  28         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  29                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  30
  31         /**#@+
  32          * @private
  33          */
  34         # Persistent:
  35         var $mLinkHooks;
  36
  37         /**#@-*/
  38
  39         /**
  40          * Constructor
  41          *
  42          * @public
  43          */
  44         function __construct( $conf = array() ) {
  45                 parent::__construct( $conf );
  46                 $this->mLinkHooks = array();
  47         }
  48
  49         /**
  50          * Do various kinds of initialisation on the first call of the parser
  51          */
  52         function firstCallInit() {
  53                 parent::__construct();
  54                 if ( !$this->mFirstCall ) {
  55                         return;
  56                 }
  57                 $this->mFirstCall = false;
  58
  59                 wfProfileIn( __METHOD__ );
  60
  61                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
  62                 CoreParserFunctions::register( $this );
  63                 CoreLinkFunctions::register( $this );
  64                 $this->initialiseVariables();
  65
  66                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
  67                 wfProfileOut( __METHOD__ );
  68         }
  69
  70         /**
  71          * Create a link hook, e.g. [[Namepsace:...|display}}
  72          * The callback function should have the form:
  73          *    function myLinkCallback( $parser, $holders, $markers,
  74          *      Title $title, $titleText, &$sortText = null, &$leadingColon = false ) { ... }
  75          *
  76          * Or with SLH_PATTERN:
  77          *    function myLinkCallback( $parser, $holders, $markers, )
  78          *      &$titleText, &$sortText = null, &$leadingColon = false ) { ... }
  79          *
  80          * The callback may either return a number of different possible values:
  81          * String) Text result of the link
  82          * True) (Treat as link) Parse the link according to normal link rules
  83          * False) (Bad link) Just output the raw wikitext (You may modify the text first)
  84          *
  85          * @public
  86          *
  87          * @param $ns Integer or String: the Namespace ID or regex pattern if SLH_PATTERN is set
  88          * @param $callback Mixed: the callback function (and object) to use
  89          * @param $flags Integer: a combination of the following flags:
  90          *     SLH_PATTERN   Use a regex link pattern rather than a namespace
  91          *
  92          * @return The old callback function for this name, if any
  93          */
  94         function setLinkHook( $ns, $callback, $flags = 0 ) {
  95                 if( $flags & SLH_PATTERN && !is_string($ns) )
  96                         throw new MWException( __METHOD__.'() expecting a regex string pattern.' );
  97                 elseif( $flags | ~SLH_PATTERN && !is_int($ns) )
  98                         throw new MWException( __METHOD__.'() expecting a namespace index.' );
  99                 $oldVal = isset( $this->mLinkHooks[$ns] ) ? $this->mLinkHooks[$ns][0] : null;
 100                 $this->mLinkHooks[$ns] = array( $callback, $flags );
 101                 return $oldVal;
 102         }
 103
 104         /**
 105          * Get all registered link hook identifiers
 106          *
 107          * @return array
 108          */
 109         function getLinkHooks() {
 110                 return array_keys( $this->mLinkHooks );
 111         }
 112
 113         /**
 114          * Process [[ ]] wikilinks
 115          * @return LinkHolderArray
 116          *
 117          * @private
 118          */
 119         function replaceInternalLinks2( &$s ) {
 120                 wfProfileIn( __METHOD__ );
 121
 122                 wfProfileIn( __METHOD__.'-setup' );
 123                 static $tc = FALSE, $titleRegex;//$e1, $e1_img;
 124                 if( !$tc ) {
 125                         # the % is needed to support urlencoded titles as well
 126                         $tc = Title::legalChars() . '#%';
 127                         # Match a link having the form [[namespace:link|alternate]]trail
 128                         //$e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
 129                         # Match cases where there is no "]]", which might still be images
 130                         //$e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
 131                         # Match a valid plain title
 132                         $titleRegex = "/^([{$tc}]+)$/sD";
 133                 }
 134
 135                 $holders = new LinkHolderArray( $this );
 136
 137                 if( is_null( $this->mTitle ) ) {
 138                         wfProfileOut( __METHOD__ );
 139                         wfProfileOut( __METHOD__.'-setup' );
 140                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
 141                 }
 142
 143                 wfProfileOut( __METHOD__.'-setup' );
 144
 145                 $offset = 0;
 146                 $offsetStack = array();
 147                 $markers = new LinkMarkerReplacer( $this, $holders, array( &$this, 'replaceInternalLinksCallback' ) );
 148                 while( true ) {
 149                         $startBracketOffset = strpos( $s, '[[', $offset );
 150                         $endBracketOffset   = strpos( $s, ']]', $offset );
 151                         # Finish when there are no more brackets
 152                         if( $startBracketOffset === false && $endBracketOffset === false ) break;
 153                         # Determine if the bracket is a starting or ending bracket
 154                         # When we find both, use the first one
 155                         elseif( $startBracketOffset !== false && $endBracketOffset !== false )
 156                              $isStart = $startBracketOffset <= $endBracketOffset;
 157                         # When we only found one, check which it is
 158                         else $isStart = $startBracketOffset !== false;
 159                         $bracketOffset = $isStart ? $startBracketOffset : $endBracketOffset;
 160                         if( $isStart ) {
 161                                 /** Opening bracket **/
 162                                 # Just push our current offset in the string onto the stack
 163                                 $offsetStack[] = $startBracketOffset;
 164                         } else {
 165                                 /** Closing bracket **/
 166                                 # Pop the start pos for our current link zone off the stack
 167                                 $startBracketOffset = array_pop($offsetStack);
 168                                 # Just to clean up the code, lets place offsets on the outer ends
 169                                 $endBracketOffset += 2;
 170
 171                                 # Only do logic if we actually have a opening bracket for this
 172                                 if( isset($startBracketOffset) ) {
 173                                         # Extract text inside the link
 174                                         @list( $titleText, $paramText ) = explode('|',
 175                                                 substr($s, $startBracketOffset+2, $endBracketOffset-$startBracketOffset-4), 2);
 176                                         # Create markers only for valid links
 177                                         if( preg_match( $titleRegex, $titleText ) ) {
 178                                                 # Store the text for the marker
 179                                                 $marker = $markers->addMarker($titleText, $paramText);
 180                                                 # Replace the current link with the marker
 181                                                 $s = substr($s,0,$startBracketOffset).
 182                                                         $marker.
 183                                                         substr($s, $endBracketOffset);
 184                                                 # We have modified $s, because of this we need to set the
 185                                                 # offset manually since the end position is different now
 186                                                 $offset = $startBracketOffset+strlen($marker);
 187                                                 continue;
 188                                         }
 189                                         # ToDo: Some LinkHooks may allow recursive links inside of
 190                                         # the link text, create a regex that also matches our
 191                                         # <!-- LINKMARKER ### --> sequence in titles
 192                                         # ToDO: Some LinkHooks use patterns rather than namespaces
 193                                         # these need to be tested at this point here
 194                                 }
 195
 196                         }
 197                         # Bump our offset to after our current bracket
 198                         $offset = $bracketOffset+2;
 199                 }
 200
 201
 202                 # Now expand our tree
 203                 wfProfileIn( __METHOD__.'-expand' );
 204                 $s = $markers->expand( $s );
 205                 wfProfileOut( __METHOD__.'-expand' );
 206
 207                 wfProfileOut( __METHOD__ );
 208                 return $holders;
 209         }
 210
 211         function replaceInternalLinksCallback( $parser, $holders, $markers, $titleText, $paramText ) {
 212                 wfProfileIn( __METHOD__ );
 213                 $wt = isset($paramText) ? "[[$titleText|$paramText]]" : "[[$titleText]]";
 214                 wfProfileIn( __METHOD__."-misc" );
 215                 # Don't allow internal links to pages containing
 216                 # PROTO: where PROTO is a valid URL protocol; these
 217                 # should be external links.
 218                 if( preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $titleText) ) {
 219                         wfProfileOut( __METHOD__ );
 220                         return $wt;
 221                 }
 222
 223                 # Make subpage if necessary
 224                 if( $this->areSubpagesAllowed() ) {
 225                         $titleText = $this->maybeDoSubpageLink( $titleText, $paramText );
 226                 }
 227
 228                 # Check for a leading colon and strip it if it is there
 229                 $leadingColon = $titleText[0] == ':';
 230                 if( $leadingColon ) $titleText = substr( $titleText, 1 );
 231
 232                 wfProfileOut( __METHOD__."-misc" );
 233                 # Make title object
 234                 wfProfileIn( __METHOD__."-title" );
 235                 $title = Title::newFromText( $this->mStripState->unstripNoWiki($titleText) );
 236                 if( !$title ) {
 237                         wfProfileOut( __METHOD__."-title" );
 238                         wfProfileOut( __METHOD__ );
 239                         return $wt;
 240                 }
 241                 $ns = $title->getNamespace();
 242                 wfProfileOut( __METHOD__."-title" );
 243
 244                 # Default for Namespaces is a default link
 245                 # ToDo: Default for patterns is plain wikitext
 246                 $return = true;
 247                 if( isset($this->mLinkHooks[$ns]) ) {
 248                         list( $callback, $flags ) = $this->mLinkHooks[$ns];
 249                         if( $flags & SLH_PATTERN ) {
 250                                 $args = array( $parser, $holders, $markers, $titleText, &$paramText, &$leadingColon );
 251                         } else {
 252                                 $args = array( $parser, $holders, $markers, $title, $titleText, &$paramText, &$leadingColon );
 253                         }
 254                         # Workaround for PHP bug 35229 and similar
 255                         if ( !is_callable( $callback ) ) {
 256                                 throw new MWException( "Tag hook for $name is not callable\n" );
 257                         }
 258                         $return = call_user_func_array( $callback, $args );
 259                 }
 260                 if( $return === true ) {
 261                         # True (treat as plain link) was returned, call the defaultLinkHook
 262                         $args = array( $parser, $holders, $markers, $title, $titleText, &$paramText, &$leadingColon );
 263                         $return = call_user_func_array( array( 'CoreLinkFunctions', 'defaultLinkHook' ), $args );
 264                 }
 265                 if( $return === false ) {
 266                         # False (no link) was returned, output plain wikitext
 267                         # Build it again as the hook is allowed to modify $paramText
 268                         return isset($paramText) ? "[[$titleText|$paramText]]" : "[[$titleText]]";
 269                 }
 270                 # Content was returned, return it
 271                 return $return;
 272         }
 273
 274 }
 275
 276 class LinkMarkerReplacer {
 277
 278         protected $markers, $nextId, $parser, $holders, $callback;
 279
 280         function __construct( $parser, $holders, $callback ) {
 281                 $this->nextId   = 0;
 282                 $this->markers  = array();
 283                 $this->parser   = $parser;
 284                 $this->holders  = $holders;
 285                 $this->callback = $callback;
 286         }
 287
 288         function addMarker($titleText, $paramText) {
 289                 $id = $this->nextId++;
 290                 $this->markers[$id] = array( $titleText, $paramText );
 291                 return "<!-- LINKMARKER $id -->";
 292         }
 293
 294         function findMarker( $string ) {
 295                 return (bool) preg_match('/<!-- LINKMARKER [0-9]+ -->/', $string );
 296         }
 297
 298         function expand( $string ) {
 299                 return StringUtils::delimiterReplaceCallback( "<!-- LINKMARKER ", " -->", array( &$this, 'callback' ), $string );
 300         }
 301
 302         function callback( $m ) {
 303                 $id = intval($m[1]);
 304                 if( !array_key_exists($id, $this->markers) ) return $m[0];
 305                 $args = $this->markers[$id];
 306                 array_unshift( $args, $this );
 307                 array_unshift( $args, $this->holders );
 308                 array_unshift( $args, $this->parser );
 309                 return call_user_func_array( $this->callback, $args );
 310         }
 311
 312 }